Monitoring in supply chain management

MELE, FERNANDO D.; MUSULIN, ESTANISLAO; PUIGJANER, LUIS

Austin, EEUU

Conferencia; AIChE Annual Meeting 2004; 2004

AIChE

<!-- /* Style Definitions */ p.MsoNormal, li.MsoNormal, div.MsoNormal {mso-style-parent:""; margin:0cm; margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:12.0pt; font-family:"Times New Roman"; mso-fareast-font-family:"Times New Roman"; mso-ansi-language:EN-GB;} @page Section1 {size:612.0pt 792.0pt; margin:70.85pt 3.0cm 70.85pt 3.0cm; mso-header-margin:36.0pt; mso-footer-margin:36.0pt; mso-paper-source:0;} div.Section1 {page:Section1;} --> This work presents an extension of multi-variated statistical methods to perform Supply Chain Monitoring (SCMo). These methods perform a dimensionality reduction that allows extracting the latent system variance from a pool of hundreds of correlated stored variables. Statistical methods such as Principal Component Analysis (PCA) and Partial Least Squares (PLS) have been implemented in the chemical industry all over the world to analyze and visualize large amounts of data. They only need past data regarding the common-cause variation of the system. In addition they can deal with missing data, ill-conditioned data, overwhelming size of data bases and the lack of good models; all of them problems that often arise when working with the Supply Chain (SC). It is important to highlight that when extending statistical methods to the SC is imperative the analysis of the type of stored data and its quality in order to account for specific issues on SCM (e.g. discrete signals, delays and don-linearities.) The proposed approach has been tested by using data generated through an event discrete simulation model. The SC scheme considered in the simulation consists of six entities connected between them. There is a supplier, a plant, two distribution centres and two markets. The model is able to use several demand models, inventory and transportation policies, which has been used to define several scenarios. Preliminary results have revealed statistical multivariate techniques as useful techniques for SCMo. It is planned to further investigate in this direction.